This paper raises the issue of speech database reduction adapted to a specific domain for Text-To-Speech (TTS) synthesis application. We evaluate several methods: a database pruning technique based on the statistical behaviour of the unit selection algorithm and a novel method based on the Kullback- Leibler divergence. The aim of the former method is to eliminate the least selected units during the synthesis of a domain specific training corpus. The aim of the latter approach is to build a reduced database whose unit distribution approximates a given target distribution. We compare the reduced databases. Finally we evaluate these methods on several objective measures given by the unit selection algorithm.
Cite as: Krul, A., Damnati, G., Yvon, F., Boidin, C., Moudenc, T. (2007) Adaptive database reduction for domain specific speech synthesis. Proc. 6th ISCA Workshop on Speech Synthesis (SSW 6), 217-222
@inproceedings{krul07_ssw,
author={Aleksandra Krul and Géraldine Damnati and François Yvon and Cédric Boidin and Thierry Moudenc},
title={{Adaptive database reduction for domain specific speech synthesis}},
year=2007,
booktitle={Proc. 6th ISCA Workshop on Speech Synthesis (SSW 6)},
pages={217--222}
}